April 2018
The demo combines the face detection from Google FaceNet and the age&gender classification from Keras. You can refer the following links for further details:
Demo video: link
- Ubuntu 16.04
- Tensorflow 1.5.0
- Keras 2.0.8
- OpenCV 3.2.0
- Python 3.5.2
Kindly note that it really matters to apply compatible version of each library to speed up the installations. I tried several combinations and came out with the above library versions.
- CPU: (TBD)
- GPU: (TBD)
- Download the FaceNet library from its offical repository and put it under the root directory.
- Run main.py
Some other issues are decribed in main.py
Issue | Solution |
---|---|
Error message: ValueError: Tensor Tensor("dense_1/Softmax:0", shape=(?, 2), dtype=float32) is not an element of this graph. | Add age_gender_model._make_predict_function() (please refer to line 38 in main.py) |
Please feel free to share the issues you meet in the comment below so that we can establish a more comprehensive Q&A list.